<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Hayden Goracke</title>
    <description>The latest articles on DEV Community by Hayden Goracke (@haydengoracke).</description>
    <link>https://dev.to/haydengoracke</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4243485%2F933a6891-a961-43ff-bc9b-f0ed25f8b2e9.png</url>
      <title>DEV Community: Hayden Goracke</title>
      <link>https://dev.to/haydengoracke</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/haydengoracke"/>
    <language>en</language>
    <item>
      <title>What broke when I put an AI voice agent on real phone lines</title>
      <dc:creator>Hayden Goracke</dc:creator>
      <pubDate>Fri, 20 Nov 2026 17:51:59 +0000</pubDate>
      <link>https://dev.to/haydengoracke/what-broke-when-i-put-an-ai-voice-agent-on-real-phone-lines-f4b</link>
      <guid>https://dev.to/haydengoracke/what-broke-when-i-put-an-ai-voice-agent-on-real-phone-lines-f4b</guid>
      <description>&lt;p&gt;I build AI phone agents for home service companies. A voice agent (Retell) answers the call, and n8n workflows do the work behind it: checking a calendar, booking, writing to the CRM. These are the failures that taught me the most, and what fixed each one.&lt;/p&gt;

&lt;h2&gt;
  
  
  1. The model was bad at "are we open right now?"
&lt;/h2&gt;

&lt;p&gt;The prompt had the business hours and the current time. The model still got the open or closed call wrong often enough to matter.&lt;/p&gt;

&lt;p&gt;The fix was to stop asking the model. An n8n step works out the answer and passes it in as a variable, and the prompt just reads the flag. My rule now: if a rule misfires twice, it leaves the prompt and gets computed upstream. Models follow data better than they follow conditional logic.&lt;/p&gt;

&lt;h2&gt;
  
  
  2. The agent said "you are booked" when nothing was booked
&lt;/h2&gt;

&lt;p&gt;The worst failure on a phone line is a caller who thinks they have an appointment and does not.&lt;/p&gt;

&lt;p&gt;Another prompt rule did not hold, so I made it structural. The booking tool now goes through a small n8n proxy that reads the transcript first. No calendar write happens unless the agent actually said a time out loud and the caller replied to it. And the agent is not allowed to say "booked" unless the tool came back with success = true.&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Dead air during tool calls
&lt;/h2&gt;

&lt;p&gt;The booking tool had a 120 second timeout, and callers were sitting through as much as 27 seconds of silence when the calendar was slow. I cut the timeout to 25 seconds and let the agent talk while the tool runs.&lt;/p&gt;

&lt;h2&gt;
  
  
  4. Publish succeeded and the phone kept running the old version
&lt;/h2&gt;

&lt;p&gt;A publish call can return success while the phone number is still bound to the previous version of the agent. Now every publish ends the same way: fetch the agent again, check which version the number is actually bound to, then place a real call.&lt;/p&gt;

&lt;h2&gt;
  
  
  5. My tests passed and production failed
&lt;/h2&gt;

&lt;p&gt;I was testing workflows with payloads I typed by hand. In production the tool arguments arrive nested differently, so my tests were passing against a shape that never shows up on a live call. I now replay real captured webhook bodies and nothing hand-built.&lt;/p&gt;

&lt;h2&gt;
  
  
  6. One caller, two bookings
&lt;/h2&gt;

&lt;p&gt;A caller rang back an hour later and got a second callback booked next to the first. The booking step now looks for an existing booking from the same number and moves it.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I would tell someone starting
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Read the actual tool call data, not the transcript. The transcript is what the model said happened.&lt;/li&gt;
&lt;li&gt;Anything that has one right answer (time, hours, dates) belongs in code.&lt;/li&gt;
&lt;li&gt;A publish is not done until a real call lands on the new version.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;This post was written with AI assistance and edited by me.&lt;/p&gt;

</description>
      <category>ai</category>
      <category>automation</category>
      <category>lowcode</category>
      <category>tutorial</category>
    </item>
  </channel>
</rss>
